skills.rest
reason-machines
reason-machinesOFFICIAL
Explore all skills by this author →
Website𝕏@reasonmachinesGitHub

multi-llm-mcp-server

Orchestrates Codex CLI and queries multiple LLM providers in parallel via MCP.

7|2|Updated May 16, 2026
Search Tags:#code review#claude code#MCP server#codex cli#FastMCP#multi-LLM#parallel query
One-click install
npx skills add https://github.com/reason-machines/mcp-skills --skill multi-llm-mcp-server-reason-machines
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: multi-llm-mcp-server
Source: https://github.com/reason-machines/mcp-skills/tree/main/skills/multi-llm-mcp-server
Command: npx skills add https://github.com/reason-machines/mcp-skills --skill multi-llm-mcp-server-reason-machines

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires fastmcp, openai.

What problem does it solve? Getting a second opinion from AI models or delegating coding tasks to Codex CLI normally requires switching between tools and manually comparing outputs. This Skill lets Claude Code query GPT, Kimi, DeepSeek, and Qwen simultaneously and offload file modifications to Codex CLI from a single interface. ## Core Features & Use Cases - Parallel Multi-Model Queries: Send the same prompt to multiple LLM providers at once and poll for aggregated responses with ask_many and wait_many. - Codex CLI Delegation: Execute code analysis, refactoring, and testing tasks through Codex CLI with read-only, workspace-write, or full-access sandbox modes. - Session Management: Maintain persistent conversation sessions per model with tools to list, clear, and isolate contexts. - Use Case: When deciding between FastAPI and Flask for an async microservice, query GPT, DeepSeek, and Kimi in parallel, then have Codex CLI scaffold the chosen framework in workspace-write mode. ## Quick Start Ask the configured models GPT, DeepSeek, and Kimi the same architecture question in parallel and compare their responses.

Frequently Asked Questions about multi-llm-mcp-server

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I query multiple LLMs at the same time from Claude Code?▼

Use the ask_many tool with a list of models such as gpt, deepseek, kimi, and qwen plus your prompt. If the job is still running, poll for completion with wait_many using the returned job_id to get all responses.

How to use Codex CLI to modify project files safely?▼

Call ask_codex with sandbox_mode set to workspace-write so Codex can only modify files within your workspace. Start with read-only mode for analysis, then escalate to workspace-write for refactoring or test generation.

Which LLM providers does this MCP server support?▼

The server supports GPT (OpenAI), DeepSeek, Kimi (Moonshot AI), and Qwen (Alibaba DashScope) out of the box. Claude is available but commented out by default and requires uncommenting the provider block and setting ANTHROPIC_API_KEY.

Why is my model API key not detected by the server?▼

Run health_check to see configured models and missing keys. Verify the environment variable is set with echo $OPENAI_API_KEY or equivalent, and restart your terminal or IDE after setting variables with setx on Windows.

What happens when a Codex or multi-model task takes too long?▼

Long-running tasks return a job_id with status running instead of blocking. Poll with wait_codex or wait_many using that job_id until the status becomes completed rather than resubmitting the request.

How do I manage conversation sessions across models?▼

Pass a descriptive session_id to the ask tool to maintain context across follow-up messages with one model. Use list_sessions to view active sessions, clear_session to remove one, or clear_all_sessions to reset everything.

Related Skills

Explore More Software Engineering Skills→
reason-machinesreason-machines
4

deepcode-cli

Configure and operate a terminal AI coding assistant for DeepSeek v4 models.

Official
Intermediate
kazuphkazuph

codex

Execute AI-driven code analysis, implementation, and review tasks via Codex CLI.

Community
Advanced
eliottbusinesseliottbusiness

requesting-code-review

Automate pre-commit code review with static analysis and security scanning.

Community
Advanced
AIsa-teamAIsa-team
23

llm-router

Route chat, vision, and comparison requests across 70+ AI models via one API key.

Official
Basic